ICCVAM Biennial Report 2024-2025

ICCVAM member agencies participate in the development and review of chemical testing guidelines issued by the Test Guidelines Programme of the Organisation for Economic Co-operation and Development (OECD) and collaborate on projects supporting the Integrated Approaches to Testing and Assessment (IATA) case studies program under the Working Party for Hazard Assessment. OECD test guidelines (TGs) and related documents are used by government, industry, and independent laboratories of the 38 OECD member countries to assess chemical safety. The U.S. National Coordinator for the OECD Test Guidelines Programme, who is a member of ICCVAM, facilitates the development, review, and clearance of a comprehensive U.S. interagency position for all draft test guidelines, guidance documents, project proposals, workshop reports, and other documents of the supported programs. The National Coordinator also coordinates expert group membership and project leadership for U.S. sponsored projects. The National Coordinator heads the delegation that represents the United States at the annual meeting of the Working Group of National Coordinators and in other test guideline development activities. One or more ICCVAM subject matter experts may join the National Coordinator at this meeting.

In 2024 and 2025, ICCVAM agencies actively participated and commented on draft OECD documents through the National Coordinator. ICCVAM members and/or NICEATM staff also supported a wide range of OECD activities during 2024 and 2025, such as:

• Supporting OECD activities to revise Test Guidelines Programme resources and processes to better support the uptake of emerging technologies, particularly for in silico methods. Efforts are ongoing to develop criteria for evaluation of in silico methods in defined approaches.
• Co-leading a project and participating in an OECD expert group coordinating revision of Guidance Document 34 on validation and acceptance of new or updated test methods for hazard assessment. ICCVAM experts participated in one in-person meeting for the drafting of the document in December 2025, in addition to ongoing virtual meetings. ICCVAM members are leading drafting of revised chapters on the essential elements of validation, the design and conduct of validation studies, and developing readiness criteria for methods undergoing validation. It is anticipated that the revised guidance document will be submitted for approval in the 2027 review cycle.
• Participating in a workshop on operation and financial aspects of validation under the OECD Test Guidelines Programme. The workshop was part of broader discussions aimed at strengthening regulatory confidence and modernizing validation processes for new and emerging chemical safety testing methods. Key points included the need for increased public funding for validation activities, particularly independent peer review, and improved visibility of existing validation resources and expertise to better support and guide method developers in facilitating validation efforts.
• Contributing to a proposal to update OECD TG 496 for in vitro test methods to identify eye irritants. TG 496 was updated in 2024 to add the OptiSafe test method.
• Leading a project to update OECD TG 497 for defined approaches to identify skin sensitizers. TG 497 was updated in 2025 to allow in vitro and in chemico methods in TGs 442C, 442D, and 442E to be used as alternate sources of information. The update also expands TG 497 to include Skin Allergy Risk Assessment–Integrated Chemical Environment (SARA-ICE), a defined approach for the determination of points-of-departure for skin sensitization potential.
• Leading a project to develop a Detailed Review Paper to “Facilitate the Development of Non-animal Test Methods to Predict the Respiratory Sensitisation Potential of Low Molecular Weight Chemicals.” The paper will support the development of an IATA for respiratory sensitization. The Detailed Review Paper was submitted to the Working Group of National Coordinators for the 2026 review cycle. An IATA project was submitted to the Working Party on Hazard Assessment for inclusion on the 2026 work plan. 
• Co-leading a project with the OECD and the European Food Safety Authority to develop an IATA framework template specifically for developmental neurotoxicity. The project will build on the existing developmental neurotoxicity IATA case studies, support the development of new ones, and provide guidance on in vitro to in vivo extrapolation, as well as standardizing uncertainty analysis and data integration within weight-of-evidence assessment. The project was approved by the Working Party on Hazard Assessment in June 2024.
• Participating in a group developing a case study on the use of an IATA for identifying systemic toxicity, which has been submitted to the OECD Working Party on Hazard Assessment.
• Serving as members and cochairs of the Developmental Neurotoxicity Expert Group. In December 2024, a “Workshop on Critical Innovations in Pesticides Safety Testing and Chemical Risk Assessment for Developmental Neurotoxicity” was convened, cochaired by U.S. experts alongside representatives from OECD and the European Food Safety Authority. The workshop examined key challenges identified through regulatory case studies, with the goal of defining priority questions and action items to be addressed in the short to medium term for regulatory implementation. Its overall aim was to support a more efficient regulatory framework by advancing the use of NAMs for assessing developmental neurotoxicity risks, particularly for chemicals and pesticides. 
• Participating in expert groups that are evaluating methods to assess thyroid disruption, skin sensitization, respiratory sensitization, and defined approaches for skin sensitization.
• Participating in an advisory group on emerging science in chemicals assessment.

OSHA is leading efforts in the U.S. to update and expand the use of nonanimal test data for the United Nations Globally Harmonized System for Classification and Labelling of Chemicals (GHS). OSHA participated in meetings convened by the United Nations Subcommittee of Experts on the GHS in 2024 and 2025 to harmonize international standards on safe handling of chemicals. OSHA held multiple public meetings to receive input from U.S. stakeholders in advance of the international meetings. The most recent hazard classes to be updated for use of nonanimal methods include skin sensitization and germ cell mutagenicity. Updates were also made to the introductory chapter that provides overarching principles to be considered when developing chemical hazard classifications. The update of the GHS chapter on skin sensitization was finalized in 2024, which included new in vitro tests and nontesting methods for classifying hazards. Guidance on the use of in silico methods and defined approaches was also included in the updated chapter. The update to the germ cell mutagenicity chapter is expected to be finalized by the end of 2026 and will include enhanced use of in vitro methods and other NAMs. Guidance of the use of these methods is being included in this update.

NICEATM scientist Kamel Mansouri, Ph.D., joined international experts from regulatory agencies, industry, academia, and federal agencies at a Spring 2025 workshop (Willett et al. 2026) to create a clear strategy for replacing animal testing with nonanimal approaches to evaluate acute toxicity. During the workshop, convened by the Animal-Free Safety Assessment (AFSA) Collaboration, participants worked on designing decision trees, step-by-step guides that help users determine the best methods for assessing the safety of chemicals without animal testing. These decision trees outline when testing can be waived, what NAMs are available, and how to ensure consistent and transparent processes across chemical regulations. They also support compliance with requirements to use animal studies only as a last resort.

The primary focus was on creating decision trees for evaluating acute oral toxicity, eye irritation and corrosion, skin irritation and corrosion, and skin sensitization. The workshop reviewed the framework's suitability for these endpoints, made recommendations for its effective implementation, and developed a timeline for completing detailed guidelines for each case.

This initiative marks a major step toward phasing out animal testing by promoting NAMs and ensuring widespread understanding and adoption of alternative methods. Future publications will address these endpoint-specific decision trees, guiding industries and regulators in adopting nonanimal strategies while maintaining reliable safety assessments for chemicals.

In the European Union, the assessment of acute toxicity is still largely based on tests conducted in live rodents. However, in July 2023, the European Commission announced plans to phase out animal testing for chemical safety assessments, with a roadmap set to be published in early 2026. A significant near-term goal is using computer-based in silico tools as the primary way to evaluate acute oral toxicity, replacing traditional animal testing. NICEATM scientists are members of an international expert group that has been working toward this goal, and their efforts show promising results. The group analyzed current European Union legal frameworks to identify regulatory requirements and highlight areas where guidance or laws may need updating to support the adoption of in silico methods (Pradeep et al. 2025). They also discussed benchmarks for how accurate in silico models need to be before they can fully replace live-animal testing. A key accomplishment includes curating a large, reliable dataset on acute oral toxicity and evaluating the performance of a specific in silico tool, the Collaborative Acute Toxicity Modeling Suite (Mansouri et al. 2021). Their results serve as a starting point for validating additional in silico methods. Looking ahead, the roadmap will explore broader opportunities to phase out animal testing. This includes studying conditions for waiving acute oral toxicity animal tests entirely, developing in vitro (laboratory-based) methods to complement in silico approaches for special cases, and creating new strategies to assess toxicity from skin and inhalation exposure without using animals. This initiative demonstrates significant progress in reducing animal testing and advancing ethical, modern tools for chemical safety evaluations while maintaining high scientific standards.

Federated learning is an approach to machine learning in which models are trained across multiple decentralized devices or nodes. Researchers at the FDA’s National Center for Toxicological Research and collaborators proposed federated learning as a practical solution to enable data-centric collaboration between regulatory agencies without requiring the direct sharing of sensitive data (Horst et al. 2025). The authors, representing Swissmedic, FDA, and the Danish Medicines Agency, suggest that current approaches to data sharing that preserve privacy (deidentification, aggregation, and synthetic data generation) each introduce trade-offs that reduce model utility. Federated learning addresses these limitations by allowing machine learning models to be trained locally at each agency and sharing only aggregated model weights rather than underlying data. This approach uses TRICIA, an existing artificial intelligence tool developed by Swissmedic, to triage and risk-score incoming serious incident reports for medical devices, as a proof-of-concept use case. A federated learning pipeline for TRICIA was deployed and tested during the 14th Global Summit on Regulatory Science using five clients from the three agencies, with synthetic data generated by GPT-4 used to simulate real incident reports. Although predictive accuracy was modest given the small synthetic dataset and limited training rounds, the federated model outperformed all individual clients, demonstrating the core principle. The paper also provides a detailed critical review of legal, security, governance, and operational considerations relevant to regulatory agencies adopting federated learning, including General Data Protection Regulation compliance, model poisoning risk, data heterogeneity across jurisdictions, intellectual property, and the need for ongoing alignment on performance metrics and model update protocols. Potential applications also extend to the development of NAMs for chemical and drug toxicity assessment, where federated learning could support the use of distributed animal-testing data across companies while protecting proprietary information and advancing the principles of replacement, reduction, and refinement of animal use.